Advances in Engineering Research, volume 191 IV International Scientific and Practical Conference 'Anthropogenic Transformation of Geospace: Nature, Economy, Society' (ATG 2019) Study of Heparanase Inhibitors (Genus: Felis) Among Bioactive Compounds Produced by Plants in Volgograd Regions

Margarita Postnova Yulia Zimina Galina Sroslova Volgograd State University, Volgograd State University, Volgograd State University, Institute of Natural Sciences, Institute of Natural Sciences, Department of Institute of Natural Sciences, Department of Department of Bioengineering and Bioengineering and Bioinformatics Bioengineering and Bioinformatics Bioinformatics Volgograd, Russia Volgograd, Russia Volgograd, Russia [email protected], [email protected], [email protected], https://orcid.org/0000-0002-1017-531X/ https://orcid.org/0000-0002-9118-7098 https://orcid.org/0000-0001-6988-6389

Alexey Bolkunov Alexander Kovalenko Volgograd State University, Volgograd State University, Institute of Natural Sciences, Institute of Natural Sciences, Department of Department of Bioengineering and Bioengineering and Bioinformatics Bioinformatics Volgograd, Russia Volgograd, Russia [email protected] [email protected] https://orcid.org/0000-0002-5517-365X

Abstract — The paper describes a search and selection of materials are easy to get that drops the production potential heparanase inhibitors. This is inhibited to expenditures. Using plant compounds as raw materials protect an organism against various diseases such as chronic allows us to avoid high expenditures for toxicity analysis kidney disease and diabetic nephropathy. These diseases are and drops the total production costs comparing with widely distributed and often lead to death. Searching low- medicines produced from synthetic compounds. For this cost plant analogs of compounds which are able to inhibit reason, we decided to search compounds in widely their target protein activity can make the treatment cheaper distributed plants: the ones which are well known for their and more efficient. The study has been carried out by means pharmaceutical effects as well as the ones which are less of Autodock Vina software to perform molecular docking commonly used in medicine. and Protein Data Bank to obtain a crystal structure. Molecular docking has computed Folate, Folate, Riboflavin Heparanase acts both at the cell surface and within its to be the most potential heparanase ligands. These ECM, regulating different processes such as cellular compounds are produced by widely distributed plants and communication, autophagy and gene transcription. known to be harmless to different species. Moreover, this enzyme is the only known mammalian endoglycosidase which is able to degrade polymeric Keywords: inhibitors, heparanase, HPSE, docking, heparan sulfates. Given the functional diversity of heparan screening, PhytoChem, Autodock Vina sulfate, its degradation by heparanase deeply affects I. INTRODUCTION important pathophysiological processes, including tumour development, inflammation and neovascularization as Anthropogenic impact on the environment affects well as progression of kidney disease [1,2]. Increased human and animals’ organisms causing various diseases expression of heparanase can be observed in numerous such as chronic kidney disease and diabetic nephropathy. malignancies, glomerular inflammation and albuminuria. These diseases are caused by disturbing biochemical Heparanase inhibition in animals protects an organism processes that leads to increased heparanase synthesis. against glomerular disease and nephropathy including the Treatment aims to inhibit this enzyme. This way today it type caused by diabetes [1]. Diabetic nephropathy often is very important to find harmless and low-cost raw appears as a diabetes complication and leads to kidney materials that can efficiently be used as a source of disease which is one of the most serious pathologies of heparanase inhibitors. this disease [3]. Diabetes mellitus is a common endocrine According to the principle of green chemistry, a disorder in animals [4]. At the end stage nephropathy process of selecting compounds and their forms in the leads to chronic kidney disease (CKD) which is another industry should aim its harm degree to be zero. Thus, common mammalian pathology. However modern today it is getting more common to search pharmaceutical medicines are quite expensive and often not very effective effects in chemical compounds produced by plants. These (lead to death with a high probability). This way, studying

Copyright © 2020 The Authors. Published by Atlantis Press SARL. This is an open access article distributed under the CC BY-NC 4.0 license -http://creativecommons.org/licenses/by-nc/4.0/. 230 Advances in Engineering Research, volume 191

the new approaches to nephropathy and CKD treatment is TABLE I. RESULTS OF THE MODELS’ LEARNING a very important problem in modern medicine. Thus, this Model AUC study is to search plant inhibitors of heparanase for CKD and diabetic nephropathy treatment. Decision 0.912 II. MATERIALS AND METHODS (MODEL) Tree

23 plants were selected and analyzed: common yarrow Random 0.98 (Achillea millefolium); chives (Allium schoenoprasum); Forest tarragon (Artemisia dracunculus); caraway (Carum carvi); coriander (Coriandrum sativum); cumin (Cuminum Naïve 0.93 cyminum); wild carrot (Daucus carota); true lavender Bayes (Lavandula angustifolia); fennel (Foeniculum vulgare); lovage (Levisticum); chamomile (Matricaria recutita); Support 0.98 Vector lemon balm (Melissa officinalis); peppermint (Mentha piperita); basil (Ocimum basilicum); marjoram (Origanum majorana); oregano (Origanum vulgare); garden parsley Then these models were used in the process of (Petroselinum crispum); rosemary (Rosmarinus ensemble forecasting to predict the activity of plant officinalis); garden sage (Salvia officinalis); summer compounds (Figure 1). savory (Satureja hortensis); lemon thyme (Thymus citriodorus); garden thyme (Thymus vulgaris), summer squash (Cucurbita-pepo). Following software and databases were used in the study: PhytoChem [5] database for the plants’ chemical composition analysis; ChEMBL and BindingDB databases to obtain data for the models’ training; AutoDock Vina [7] for molecular docking; the Blastn algorithm [8] to analyze the homology of human and genus Felis’s heparanase sequences.

III. RESULTS AND DISCUSSION Chemical composition analysis was carried out for Fig. 1. Models of plant compounds activity in Volgograd Region each plant. 826 unique compounds were obtained. At the first step the models selected 10 compounds out In order to reduce a set of plants for experiments, we of 826 whose properties were similar to known active chose the QSAR (Quantitative structure-activity target inhibitors (Table 2). relationship) commonly used methods of virtual screening as the first step of selection and molecular docking as the TABLE II. RESULTS OF NATURAL PRODUCTS FORECASTING second one. The QSAR principle is based on the trained (FORECASTED BY ALL THE MODELS). model’s ability to relate a structure and properties of a chemical compound [6]. Thus, structural features of Chemical Organ Molar Plant species compound mass, known heparanase inhibitors would allow us to indicate g/mol new compounds having the same features and therefore PECTIN Root 194.139 Daucus-carota the same properties. Due to train the model, we selected compounds with experimental values of affinity ki, kd, IC50, EC50 for a human. The compounds were obtained PECTIN Fruit 194.139 Foeniculum-vulgare from ChEMBL (206 units) and BindingDB (231 units) databases. Salts had been removed from the set before PECTIN Seed 194.139 Cucurbita-pepo selection. The first set was divided into two groups: active and inactive. Compounds which ki, kd, IC50 and EC50 PECTIN Fruit 194.139 Coriandrum-sativum values were less or equal to 1000 nmol were considered as active. Compounds which affinity values were over or equal to 100 000 nmol were considered as inactive. PECTIN Shoot 194.139 Salvia-officinalis Middle as well as duplicate values were removed. Thus, the final set involved 188 active and 31 inactive PECTIN Shoot 194.139 Rosmarinus- compounds. These compounds were selected to train the officinalis machine learning models which are often used in virtual PECTIN Shoot 194.139 Mentha-x piperita screening. For each of the models we computed an AUC statistic metric which evaluates a quality of machine learning (5-fold cross validation) (Table 1). RIBOFLAVIN Plant 376.369 Thymus-vulgaris

RIBOFLAVIN Seed 376.369 Carum-carvi

231 Advances in Engineering Research, volume 191

Chemical Organ Molar Plant species Chemical Organ Molar Plant species compound mass, compound mass, g/mol g/mol RIBOFLAVIN Root 376.369 Daucus-carota ASPARAGINE Root 132.119 Artemisia- dracunculus

RIBOFLAVIN Leaf 376.369 Salvia-officinalis HISTIDINE Leaf 155.157 Ocimum-basilicum

RIBOFLAVIN Plant 376.369 Origanum-majorana HISTIDINE Seed 155.157 Cucurbita-pepo

RIBOFLAVIN Leaf 376.369 Mentha-x piperita HISTIDINE Fruit 155.157 Foeniculum-vulgare

RIBOFLAVIN Plant 376.369 Achillea-millefolium HISTIDINE Plant 155.157 Achillea-millefolium

RIBOFLAVIN Seed 376.369 Cucurbita-pepo HISTIDINE Leaf 155.157 Rosmarinus- officinalis

RIBOFLAVIN Plant 376.369 Artemisia- HISTIDINE Leaf 155.157 Mentha-x piperita dracunculus RIBOFLAVIN Leaf 376.369 Coriandrum-sativum The next step was molecular docking which computed RIBOFLAVIN Seed 376.369 Cuminum-cyminum the energy of binding a compound to its target. It was performed by means of the Autodock Vina software. A RIBOFLAVIN Plant 376.369 Rosmarinus- crystal structure was obtained from Protein Data Bank. In officinalis the process of molecular modeling, we bounded selected compounds with a human heparanase 5E8M (a non- RIBOFLAVIN Plant 376.369 Petroselinum-crispum mutated, native form of a protein not bounded with ). The of the protein consists of five RIBOFLAVIN Seed 376.369 Foeniculum-vulgare amino acids [9]. After QSAR models selected some compounds, their binding energy was computed by the RIBOFLAVIN Leaf 376.369 Ocimum-basilicum molecular docking method (Table 3). Docking was carried out in two methods: under the complete fixation of target’s atoms (DS_rig) and limited flexibility of the RIBOFLAVIN Leaf 376.369 Allium- active center’s atoms (DS_flex) [10]. Results of modeling schoenoprasum binding SAR hits to the heparanase active center are FOLACIN Plant 441.404 Achillea-millefolium presented in Table 3.

TABLE III. RESULTS OF MODELING BINDING SAR HITS TO THE FOLACIN Plant 441.404 Petroselinum-crispum HEPARANASE ACTIVE CENTER DS_flex, DS_rig, Kd_ Chemical Plant species FOLATE Leaf 441.404 Allium- kcal/mol kcal/mol camputed compound schoenoprasum -5.7 -5.0 ASPARAG Achillea- FOLATE Leaf 441.404 Rosmarinus- INE millefolium, officinalis Salvia- officinalis, FOLATE Leaf 441.404 Mentha-x piperita -6.9 -5.9 CODECAR Cucurbita-pepo BOXYLAS E PHEOPHYTIN-A Plant 871.22 Cucurbita-pepo -9.4 -8.9 FOLACIN Achillea- millefolium, Petroselinum- CODECARBOX Plant 247.143 Cucurbita-pepo crispum, YLASE -9.6 -9.0 FOLATE Allium- GAMMA- Plant 290.334 Allium- schoenoprasum, GLUTAMYL-S- schoenoprasum Rosmarinus- ALLYLCYSTEIN officinalis, E Mentha-x MALEIC-ACID Root 116.072 Levisticum-officinale piperita -6.8 -6.1 GAMMA- Allium- GLUTAM schoenoprasum ASPARAGINE Plant 132.119 Achillea-millefolium YL-S- ALLYLCY STEINE ASPARAGINE Plant 132.119 Salvia-officinalis -5.9 -5.3 HISTIDIN Ocimum- E basilicum, Cucurbita-pepo,

232 Advances in Engineering Research, volume 191

DS_flex, DS_rig, Kd_ Chemical Plant species problem can be solved by searching plant inhibitors of kcal/mol kcal/mol camputed compound heparanase which is a target of CKD and nephropathy. It Foeniculum- can make treatment cheaper and more efficient. vulgare, Achillea- 3 plants out of 23 have been selected to have potential millefolium, heparanase inhibitors: Folate (DS -9,6), Folacine (DS - Zingiber- 9,4), Riboflavin (DS -8,3). These compounds are officinale, produced by widely distributed plants and known to be Rosmarinus- officinalis, harmless to different species. Mentha-x piperita, Carum- REFERENCES carvi, Allium- [1] T.J. Rabelink, and B.M. van den Berg, “Heparanase: roles in cell schoenoprasum, survival, remodelling and the development of Coriandrum- kidney disease,” in Nature Reviews Nephrology, vol. 13, no. 4, pp. sativum 201-212, 2017. -5.1 -4.5 MALEIC- Levisticum- [2] B. Heyman, and Y. Yang, “Mechanisms of heparanase inhibitors in ACID officinale cancer therapy,” in Experimental Hematology, vol. 44, no. 11, pp. 1002-1012, 2016. -6.8 -6.1 PECTIN Foeniculum- vulgare, [3] I.E. Smirnov, A.G. Kucherenko, and G.I. Smirnova, “Diabetic Cucurbita-pepo, nephropathy,” in Russian Pediatric Journal, vol. 18, no. 4, pp. 43- Coriandrum- 50, 2015. sativum, Salvia- [4] C.A. Bloom, and J.S. Rand, “Feline diabetes mellitus: clinical use officinalis, of long-acting glargine and detemir,” in Journal of Feline Medicine Rosmarinus- and Surgery, vol. 16, no. 3, pp. 205-215, 2014. officinalis, [5] “Dr. Duke's Phytochemical and Ethnobotanical Databases”. URL: Myristica- https://phytochem. nal.usda.gov/phytochem/help/index/about. fragrans, [6] A. Cherkasov, and E.N. Muratov, “QSAR modeling: where have Mentha-x you been? Where are you going to?” in Journal of Medicinal piperita Chemistry, vol. 57, no. 12, pp. 4977-5010, 2014. -7.9 -5.9 PHEOPHY Cucurbita-pepo [7] O. Trott, and A.J. Olson, “AutoDock Vina: improving the speed TIN-A and accuracy of docking with a new scoring function, efficient optimization and multithreading,” in Journal of Computational -8.3 -8.2 RIBOFLA Thymus- Chemistry, vol. 31, no. 2, pp. 455-461, 2010. VIN vulgaris, Carum-carvi, [8] “Blastn”. URL: Salvia- https://blast.ncbi.nlm.nih.gov/Blast.cgi?LINK_LOC=blasthome&P officinalis, AGE_TYPE=BlastSearch&PROGRAM=blastn. Origanum- [9] “UniProtKB - Q9Y251 (HPSE_HUMAN)”, URL: majorana, https://www.uniprot.org/uniprot/Q9Y251. Mentha-x [10] A.A. Glushko, A.S. Chiryapkin, V.S. Chiryapkin, A.M. piperita, Murtuzalieva, and Yu.A. Polkovnikova, “Development of a Achillea- methodology for modeling the interaction of biologically active millefolium, substances with the active center of angiotensin converting Cucurbita-pepo, enzyme”, in Journal of Pharmacy and Pharmacology, vol. 5, no. 5, Artemisia- pp. 487-503, 2017. dracunculus, Coriandrum- sativum, Cuminum- cyminum, Rosmarinus- officinalis, Petroselinum- crispum, Foeniculum- vulgare, Ocimum- basilicum, Allium- schoenoprasum Molecular docking has selected Folate, Folacine, Riboflavin to be the most potential ligands for heparanase (binding energy is less -8 kcal/mol).

IV. CONCLUSION Heparanase enzyme plays an important role in various living processes, including cellular communication, gene transcription and autophagy. At the same time, its increased expression is observed in albuminuria, diabetic nephropathy and chronic kidney disease. However, there is no effective and low-cost medicine to treat them. This

233